Customer analytics: segmentation and lifetime value

Customer analytics shows which customers are worth keeping, which have stopped buying and what each kind is worth after returns. You get customer segmentation on readable rules for recency, frequency and value, and customer lifetime value per acquisition cohort after acquisition cost.

Last reviewed

RFM segments, example

Champions are 10% of customers and 41% of revenue

Example data
  • Champions

    10% share of customers

    41% share of revenue

    2,000£984,000

    Rule: last order within 90 days, 4 or more orders and £300 or more in 12 months
  • Loyal

    17% share of customers

    26% share of revenue

    3,400£624,000

  • New

    15% share of customers

    9% share of revenue

    3,000£216,000

  • At risk

    23% share of customers

    17% share of revenue

    4,600£408,000

  • Lapsed

    35% share of customers

    7% share of revenue

    7,000£168,000

Revenue after returns, 12 months to 31 October 2026
Example data: an invented shop with 20,000 customers who ordered in the 12 months to 31 October 2026, and £2.4m of revenue after returns. Every customer sits in exactly one segment, by the written rules in the segment rulebook.
Show the numbers
RFM segments, example data
SegmentCustomersShare of customersRevenue after returnsShare of revenue
Champions2,00010%£984,00041%
Loyal3,40017%£624,00026%
New3,00015%£216,0009%
At risk4,60023%£408,00017%
Lapsed7,00035%£168,0007%
Total20,000100%£2,400,000100%

What customer segmentation fixes

Our lifetime value is one numberOne figure, worked out on revenue, covers the whole base, and every channel's cost per customer is judged against it.
Nobody can explain the segmentsA model sorted customers into clusters two years ago, and the team can't say why a customer moved last month.
Everyone gets the same emailLast week's buyer and someone who last bought in March get the same offer on the same day.
We don't know who has stopped buyingThere's no count of how many of last year's repeat customers have gone quiet.

Cohort analysis, segments and lifetime value: what you get

The segment rulebookEach segment is a written rule on recency, frequency and value, like the Champions rule in the exhibit, so marketing can rebuild it in your email tool.
Lifetime value by cohortFor each month of first order, it shows cumulative contribution per customer at 3, 6, 12 and 24 months, after acquisition cost. Contribution is revenue after returns less cost of goods, delivery and payment fees, defined like each figure in the KPI library.
Cohort retentionOne line per monthly cohort shows the share who ordered again by month 1, 2, 3 and on.
The action sheetEach segment names one action and how it's measured, such as a win-back email at day 120 for At risk customers, with a random 10% held back.

A customer who cost £45 to win and sends back half of what they order has a lifetime value, and it can be below zero.

How a customer analytics project runs

  1. 1

    Scoping

    Scoping takes the order, refund and customer tables and ad spend by channel, and matches guest orders on email where your terms allow it.

  2. 2

    Build

    The segments are SQL rules in your warehouse, tested so every customer lands in one segment and revenue sums to the order system's total.

  3. 3

    Handover

    Marketing learns to rebuild a segment from its rule.

  4. 4

    After handover

    The segments refresh on a schedule, and moving a threshold, say Champions from four orders to five, is a one-line change.

How customer lifetime value misleads

One average for everyoneA blended figure makes a cheap channel and an expensive one look equally good, so lifetime value is worked out by cohort, and by channel where spend allows.
Churn with no cancel buttonA shop has no cancellation event, so the lapsed threshold comes from your data, at the number of days within which 90% of second orders arrive.
Segments that change nothingEvery segment on the action sheet names its action and the holdout that measures it.

In the example data, a Champion brought in £492 of revenue after returns over the year, and a Lapsed customer £24.

Customer analytics by industry

EcommerceRFM runs on orders after returns, and lifetime value by first-order month and channel.
Subscription and mediaChurn is read for each start-month cohort at every renewal, and value is counted in paid months.
SaaSSeats and plan changes move revenue inside the month, so each sign-up cohort gets gross and net revenue retention side by side.
Wholesale and B2BMargin per account is worked out after rebates, delivery and credit, and an account is flagged as lapsing when an order runs late against its usual gap.

Whether a win-back email caused the orders it claims is a marketing measurement question, and a holdout answers it. Forecasting can turn the cohorts' retention curves into next quarter's repeat orders.

Questions about segments and lifetime value

What is RFM analysis?

RFM sorts customers on recency (days since the last order), frequency (orders in a set window) and monetary value (what they spent in it). A customer who ordered 40 days ago, five times this year, for £520, lands in Champions (last order within 90 days, 4 or more orders, £300 or more in 12 months).

How do you work out customer lifetime value?

It's worked out by acquisition cohort, meaning everyone whose first order fell in the same month, as cumulative contribution per customer at 3, 6, 12 and 24 months, less acquisition cost per customer.

Do you do customer segmentation for ecommerce?

Yes. Segments are built on orders after returns, so a customer who sends back most of what they buy doesn't count as a Champion, however often they order.

Can you do churn analysis for a subscription business?

Yes. Churn is read for each start-month cohort at every renewal, and where cancellations carry a reason, voluntary cancellations are split from failed payments too.

Can you build lead scoring for a B2B sales team?

Yes, on the same principle. The score is written as rules learned from past leads that did and didn't become deals, and tested on a held-back batch before your sales team uses it.

How much order history do you have?

On the call we look at where your order history lives and say which segments and cohorts it can support.

Book a call

Free 30 min call